摘要: The AMBR project has a worthy and lofty goal—to create systems that can successfully simulate human behavior. In service of this goal, AMBR teams created category learning systems that were embedded within a cognitive architecture. To test and more fully develop these architectural proposals, human performance data were collected in a behavioral study that paired structures utilized in Shepard, Hovland, and _] enkins’s (1961) classic category leaming studies with secondary tasks that evoked air traffic control (ATC) scenarios.More narrowly defined models developed in the category learning litera-ture have successfully accounted for human performance on Shepard, Hovland, and] enkins’s (1961) six problem types (eg, Kruschke, 1992; Love 8: Medin, 1998; Nosofsky, Palmeri, 8c McKinley, 1994). One interesting question is how these task-specific models compare to the architectural solutions proposed by …